We propose Ano-SuPs, a two-stage strategy anomaly detection method for manufactured products. It reconstructs input images twice to identify suspected patches, enhancing anomaly detection accuracy. The method is validated through simulation experiments and case studies.